Presentation 2003/3/8
Decision-tree Induction from Time-series Data Based on Dynamic Time Warping
Yuu YAMADA, Einoshin SUZUKI, Hideto YOKOI, Katsuhiko TAKABAYASHI,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) This paper proposes a novel approach for learning a decision tree from a data set with time-series attributes. A time-series attribute takes, as its value, a sequence of values each of which is associated with a time stamp, and can be considered as important since it frequently appears in real-world applications. Our time-series tree has a time sequence in its internal node, and splits examples based on similarities between a pair of time sequences. We first define our standard example split test based on dynamic time warping, then propose a decision tree induction procedure for the split test. Experimental results confirm that our induction method, unlike other methods, constructs comprehensive and accurate decision trees. Moreover, a medical application shows that our time-series tree is promising in knowledge discovery.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Time-series Decision Tree / Time-series Data / Dynamic Time Warping / Knowledge Discovery / Medical Test Data
Paper # AI2002-85
Date of Issue

Conference Information
Committee AI
Conference Date 2003/3/8(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Artificial Intelligence and Knowledge-Based Processing (AI)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Decision-tree Induction from Time-series Data Based on Dynamic Time Warping
Sub Title (in English)
Keyword(1) Time-series Decision Tree
Keyword(2) Time-series Data
Keyword(3) Dynamic Time Warping
Keyword(4) Knowledge Discovery
Keyword(5) Medical Test Data
1st Author's Name Yuu YAMADA
1st Author's Affiliation Faculty of Engineering, Yokohama National University()
2nd Author's Name Einoshin SUZUKI
2nd Author's Affiliation Faculty of Engineering, Yokohama National University
3rd Author's Name Hideto YOKOI
3rd Author's Affiliation Division of Medical Informatics, Chiba University Hospital
4th Author's Name Katsuhiko TAKABAYASHI
4th Author's Affiliation Division of Medical Informatics, Chiba University Hospital
Date 2003/3/8
Paper # AI2002-85
Volume (vol) vol.102
Number (no) 711
Page pp.pp.-
#Pages 6
Date of Issue